Is there an immunohistochemical technique definitively valid in epidermal growth factor receptor assessment?

Immunohistochemistry (IHC) is required for the selection of patients for a monoclonal antibody based targeted treatment with C225 (Erbitux). To validate the usefulness of IHC, the confirmation of assays and scoring systems are mandatory. In an attempt to standardize the immunohistochemical detection of the epidermal growth factor receptor (EGFR), we retrospectively evaluated three commercially available EGFR kits or antibodies and analyzed the discrepancies between the tests in terms of the percentage of positive cells, intensity, cut-off value and fixatives. We extracted 232 paraffin-embedded samples from a metastatic colorectal cancer clinical trial. For all the cases, EGFR expression was assessed with the FDA approved Dako EGFR pharmaDx kit, the Zymed EGFR kit and the Ventana EGFR 3C6 antibody. Different cut-off values were tested, and the intensity was scored 0, 1+, 2+, 3+ following Dako's recommendations. The percentage of positive cases varied from 93 to 75% with a cut-off of value 1% of positive cells, from 80 to 61% with a cut-off of 5% positive cells and from 72% to 48% with a cut-off value of 10%. Both Ventana and Zymed tests were more sensitive than the Dako test (Ventana >Dako; p<10(-7), Zymed >Dako; p=2.10(-6)). No difference was noted between Ventana and Zymed tests (p=0.75). A high concordance was observed for the 3 tests for the evaluation of high intensities. The use of a scoring system combining the percentage of positive cells and intensity was not useful for Zymed and Ventana as the intensity of staining is correlated to the percentage of positive cells: Ventana (p<10(-6)) and Zymed (p<10(-5)). No interaction with staining was identified for any of the fixatives, or with the nature of samples received (i.e. slides vs blocks, biopsies vs surgical specimens). Our data showed a higher percentage of positive cells detected by Ventana and Zymed tests, whatever the cut-off value for positivity. No scoring system showed, to date, its accuracy, and more studies have to be conducted with an evaluation of the response to cetuximab, possibly with a correlation with FISH amplification in colorectal carcinoma.